Any student of finance understands that one of the core themes of finance is about risk and rewards. This understanding is not something limited to the academic world, but is also inherently understood by financial market participants. Since centuries ago, financial market participants have been trying hard to exploit risk to reap greater riches in the financial markets (Malkiel, 2007). However, for many decades, the ideas of risk and rewards are not something readily quantified; or more precisely, there are simply no reasonable and solid analytical tools developed to quantify risk, and the respective rewards to be expected from assuming the risks.
This is no longer the situation when Nobel Prize winners, such as William Sharpe and Harry Markowitz developed theories and framework that gave rise to modern portfolio theory. In the modern portfolio theory, perhaps the most controversial topic being discussed by academicians and scholars are the framework of Capital Asset Pricing Model (CAPM). Until today, many academicians still believe that CAPM is the center-piece of modern finance. It is popular because it provides insight and theoretically based framework in the investigation of the risk of an asset and its expected return. Besides, until present, the logic of CAPM is compelling, and up to now, many more sophisticated models in the context of security pricing are all relying heavily on the distinction between systematic and unsystematic risks (Bodie, Kane & Marcus, 2007). Generally speaking, the fundamental logic behind CAPM is that investors cannot yield abnormal return by assuming risks that can be diversified away (i.e., often, also referred as the firm-specific risks or as the unsystematic risks) through holding a perfectly diversified portfolio. In order for investors to yield higher returns, the only way is to assume risks that cannot be diversified away (i.e., often, also regarded as the market risks or the systematic risks). As such, under the CAPM theory, investors can only possibly yield above average return (i.e., better return than the average market return) by holding a portfolio of higher beta (Malkiel, 2007).
Until today, beta is still a popular concept taught in finance and business schools, despite being a highly controversial figure and concepts hotly debated by both practitioners and scholars. Conceptually, beta is the numerical representation of systematic risks (in which systematic risks also referred to as the market risks, because it is about the reaction of individual stocks to general market movement). For once, as discussed by Malkiel (2007), beta is the state-of-the-art concepts in modern finance theory, as it is the number computable to measure the subjective feelings investors are having for many years (due to the movement of a stock to the movement of the market). Generally speaking, it is understood by students studying finance that high beta stocks are those risky stocks (more appropriately, named as the aggressive stocks) whereby low beta stocks are those less risky stocks (i.e., defensive stocks). Overall, the key takeaways of the CAPM theory, in the context of risk and rewards (i.e., ultimately, the fundamental ideas about finance) is that investors will not be rewarded for bearing unsystematic risks. In other words, returns for any stocks, will be related to beta (i.e., systematic risks) of that particular stocks. It is by assuming the systematic risks, by buying higher beta stocks, which investors may be able to yield above average return from the financial market.
There are many critiques by scholars on the practicality and accuracy of CAPM. This is nothing surprising, as many studies have been repeatedly found that the cross section stock returns in many of the countries around the world, has little, or more precisely non-statistically significant relationship with the beta coefficient computed from the CAPM theory (Drew, Naughton & Veeraragavan, 2005). In other words, the beta coefficients computed from the CAPM theory has little predictive power in predicting the cross section average stock returns in real financial market. As such, it is reasonable to expect improvement to the model is necessary. In the following section, the inherent limitations of CAPM will be discussed. The improvements suggested by scholars are also discussed. Effectively, more discussion will be concentrate on a better framework, namely the multifactor models widely discussed and applied by scholars today, to understand the risk and rewards nature of the financial market.
Although CAPM is a ground-breaking concepts, persuasively arguing that the expected returns of common stocks should be linearly related to the market beta of that particular common stock, the theory hardly withstand empirical testing by scholars. Precisely, Fama and French (1992) found that the cross section of average stock returns in Wall Street (of United States) exhibit little (i.e., non-statistically significant or meaningful) relationship to the beta coefficient in CAPM. Such findings suggested that CAPM may not be practical in explaining the risk and rewards nature of common stocks in real life. Worst, Fama and French (1992) found that variables such as firm size and book-to-market equity unexpectedly able to provide statistically significant explanation on the cross section variation of average stock returns . In other words, once the researcher control for certain variables, such as the (a) size of the firm, and (b) the market value to book value ratio, the beta of a particular stock, indeed contribute nothing to the explanation of the prediction of future returns. Thus, put it bluntly, it is apparent that beta is not a good predictors of stock returns – and it cannot be used for guiding investment decisions successfully in real world. Perhaps, from a more optimistic view points, it is also clear beta alone does not able to explain the whole story of risks (Bodie, Kane & Marcus, 2007). It is not reflective in suggesting the expected returns to the relevant risks of common stocks in the financial markets.
As beta is found to not able to explain securities’ risk factor satisfactorily, it is argued by researchers that there are risk factors that are affecting a particular common stock expected returns beyond the beta calculation. As argued by Drew et. al. (2005), it is apparent that beta one dimensional measure of risks is not sufficient. Indeed, Fama and French (1992) argued that if stocks are priced in a rational manner, the relevant risks pertaining to these stocks should be multi-dimensional. In their published journals, the two dimensions of risks found to be relevant and capable of explaining variation in cross section average stock returns include: (a) the size of the firm, and (b) the market value to book value ratio. Thus, following the findings presented by Fama and French (1992), it is then suggestive that there are other sources of systematic risks not being captured by CAPM beta. Indeed, the findings of Fama and French (1992) are popularly referred to as the three factor model, whereby the model can be written mathematically as follow:
Rpt – Rft = apt + bp (Rmt-Rft) + spSMBt + hpHMLt + ept
|Rpt||=||Average returns of a certain portfolio|
|Rft||=||Risk free rate|
|Rmt||=||Equal weight market return|
|SMBt||=||Monthly difference between the return on a portfolio of small stocks and a portfolio of big stocks|
|HMLt||=||Monthly difference between the return on a portfolio of high book-to-market equity stocks and the return on a portfolio of low book-to-market equity stocks|
Thus, from the equation stated above, the three factors used to explain cross section average stock returns are: (a) beta (i.e., the tendency of a particular stock to move more or less than the market), (b) the size effect, and lastly, (c) the book to market ratio. Aside from beta, the other two factors were added to explain cross section average stock returns because it is observed that the average returns from smaller stocks tend to outperformed those of the larger stocks historically. Secondly, it is also observed that stocks traded with higher ratio of book value to market value ratio tend to outperform those stocks with lower book value to market value ratio historically. There are many economically sensible reasons to explain such phenomenon. Firstly, as discussed and asserted by Fama and French (1992), firms with higher book value to market value ratio are more likely to be in a financially distressed position, and thus, these securities should have higher expected returns, to compensate for the higher risks associated with them. Then, the authors also argued that smaller size firms tend to more susceptible to changes in business cycles, and thus, more risky. As such, when the stocks of smaller size or the stocks with higher book value to market value ratio are compensated more if compared to the other stocks, it is an indication that the macroeconomic risk factors on these stocks are higher.
Apart from the three factors suggested by Fama and French (1992), there are many other relevant systematic risks found by researchers to be statistically meaningful and significant in affecting cross section average stock returns in different countries around the world. As argued by Malkiel (2007), it is rational to think that other systematic risk elements should be included to the one factor CAPM, to better explain the variation of stock returns. He asserted that some of the systematic risk elements found to be statistically significant in predicting stock returns include the following: (a) changes in national income; (b) changes in interest rates; (c) changes in inflation; and (d) the extent of disagreement among security analyst’s forecasts for each individual stock. He argued that national income is a good proxy of personal disposable income. As individual disposable income increases, they may spend more, and thus, drive up the corporate profitability. Hence, the level of changes in national income should be rationally expected to affect the stock returns in a systematic manner. Similarly, the changes of interest rates should affect the various businesses in a systematic manner as well. When interest rates increase, the costs of doing business as well as the cost of investment tend to increases, making the corporations and firms harder to make more profitable investments. Besides, as interest rates increases, people tend to defer buying those big items and hence, put downward pressures on the demands of the expensive goods. As such, when the demand in the economy reduced, the business prospects of companies may not be as good as when the case the interest rates are low. Hence, interest rates should be one of the systematic factors affecting stock returns, and should be included in the multifactor model. Then, the changes of inflation rate will also have impacts to the business landscape. When inflation is too high, people may defer spending, or cut down spending on unnecessary or luxury items. This will likely to put downwards pressure to the economy, and hence, affecting the profitability of the firms adversely. Lastly, and most interestingly, Malkiel (2007) also pointed out that the extent of disagreement among security analysts’ forecasts for each individual stock is perhaps the most predictive variable that can be used in a multifactor model. This is because the diversity of analysts’ opinion on a particular stock indicate that the prospect of the stock is highly debatable, and thus, more risky.
However, there is no authoritative study that can confirm which are the most powerful and predictive variables to be used to represent the systematic factors in a multifactor model. Thus, the best approach to understand the current stages of research in the context of multifactor models is to study the existing different types of multifactor models suggested by different scholars. With that, readers can get a sense of the elements or factors often being priced in to represent the systematic forces. Different types of multifactor models will be discussed in the following section.
Generally speaking, the various types of multifactor model can be divided into two main categories, namely: macroeconomic and microeconomic multifactor model. Macroeconomic multifactor model are those multifactor model developed through incorporating macroeconomic factors, such as real Gross Domestic Products growth, unexpected inflation, interest rates, or other monetary variables to explain stock returns. These macroeconomic factors are perceived as influential towards affecting most of the expected cash flow of the many companies, and hence, are believed to be able to affect stock prices. Then, similarly, microeconomic multifactor models employ microeconomic factors as the risk factors to explain variation in stock returns. Microeconomic factors are those risk factors that can be measured by the specific characteristics of the firm themselves. Some example of microeconomic factors often used include: price-to-earnings ratio, market capitalization of a firm (i.e., firm size), book-to-market value ratio, and dividend yield (Connor, 1995).
There are various different macroeconomic multifactor models being developed by scholars in attempting to explain stock returns. Some of the famous one will be discussed in this section. In this context, Chen, Roll and Ross model is one of the earliest macroeconomic multifactor models developed (Bodie, Kane & Marcus, 2007). In the model, a total of six factors are used to explain stock returns. These six factors include: (a) return on a value weighted index of NYSE-listed stocks, (b) monthly growth in US industrial production, (c) changes in monthly inflation rate; proxy by Consumer Price Index (CPI) in US, (d) unexpected inflation (as computed as the differences between actual inflation rate to unexpected inflation rate), (e) unexpected bond credit spread changes, and (f) unexpected shift in term structure (Chen, Roll & Ross, 1986). Then, another popular macroeconomic multifactor model is the Burmeister, Roll and Ross model. In this model, five factors are used to explain variation in stock returns. These five factors are: (a) confidence risk, (b) time horizon risk, (c) inflation risk, (d) business cycle risk, and (e) market timing risk (Fabozzi, Focardi & Kolm, 2006).
Then, the famous and popularly researched microeconomic multifactor models include Fama and French three factor model as well as Carhart model (Connor, 1995). As will be discussed extensively in sections below, the three microeconomic factors employed under the Fama and French three factor model include: (a) excess return on market portfolio, (b) return on a small-cap portfolio minus the return on a large-cap portfolio, and (c) return on a portfolio of high book-to-price stocks minus the return on a portfolio of low book-to-price stocks (Bodie, Kane & Marcus, 2007). Then, Carhart model add another extra factor to the famous three factors just mentioned above. The fourth factor added is a momentum factor, as calculated by the return on the portfolio of best performing stocks minus the return on the portfolio of worst performing stock (Fabozzi, Focardi & Kolm, 2006).
Realizing that the theory of CAPM has many inherent limitations, and that theory is often proven not able to withstand empirical tests, researchers have been switching to other models to investigate the risk and returns of assets traded in financial markets. Following the spirit of CAPM, it is widely argued that improvement to the theory of CAPM can be performed by developing models that incorporate multiple sources of risks relevant to the market. Essentially, that gave rise to many different versions of multifactor models of risks and returns. Indeed, many of the models suggested by different researchers indeed produce encouraging results, whereby some of the models able to explain the cross section average returns of stocks better than the CAPM theory. For examples, as pointed out by Drew et. al. (2005), studies had found that incorporation of extra variables, such as firm size, earnings to price ratio, book value to market value ratio and idiosyncratic volatility, in the investigation of cross section average stock returns in the United States able to explain stock returns better than the traditional beta coefficient suggested by CAPM. In the next few paragraphs, several studies on this context will be presented.
Drew et. al. (2005) had conducted a study to investigate the priced risk factors in Shanghai Stock Exchange. Consistent with the theory there are systematic factors (i.e., market factors) are affecting the individuals stock returns, the researchers able to find evidences supporting such notion. Specifically, it is found that overall market factors explain a total of 61% of the variation in the cross section of average stock returns in Shanghai Stock Exchange. The authors found that beta coefficients are statistically significant at the 1% level. Not only is that, the researchers also able to find evidences that smaller size stocks as well as those stocks with low book value to market value ratio able to provide higher returns when compared to those stocks of bigger size or those of high book value to market value ratio. The authors concluded that their findings are similar to the findings from Fama and French (1992) in the sense that smaller stocks in Shanghai Stock Exchange under the period of investigation tend to exhibit better returns. It is argued that such a phenomenon present because smaller stocks are generally riskier, and thus, carry risk premium. However, their findings that stocks with low book value to market value ratio outperformed those stocks with high book value to market value ratio is not something consistent to the findings of Fama and French (1992). The authors subsequently argued that the book to market value ratio is not as pervasive as suggested by Fama and French (1992). Nonetheless, as the R2 of the three factors regression models in their study is 0.91, as compared to the R2 of the single factor CAPM regression models (i.e., beta as sole explanatory variables) of 0.61, it is asserted that the three factors model is better in the sense that it able to explain more variation in average stock returns than the one factor CAPM. Overall, their studies provide evidences that there are other sources of risks in the systematic factors, and that is true in other countries such as China.
The notion that multifactor models are able to explain the variation in the cross section of average stock returns is also found to be true in other country. For example, a study conducted by Javid (2008) to test the conditional CAPM to conditional multifactor CAPM within individual stocks in Karachi Stock Exchange, Pakistan, from the year 1993 to 2004, found that the conditional multi-factor CAPM can explain the variation of cross section average stock returns better than the conditional (one factor) CAPM. This hold true even in a out-of-sample tests conducted by the authors. In the study conducted by Javid (2008), the other factors included in the multifactor CAPM include macroeconomic variables that are related to fluctuation in business cycles, such as the market returns, the term structure, the growth of industrial index, inflation rate, the foreign exchange rate as well as the changes in oil prices. Following their findings, the authors argued that economic variables used in their study are proven to provide meaningful insights for investors to predict stock returns in the future. Aside from the Karachi Stock Exchange in Pakistan, Naughton and Veeraraghavan (2005) had also conducted a study on the validity and explanatory power of three factors models in three countries as follow: Indonesia, Singapore and Taiwan. Their findings provide empirical evidences that the risk premium for firm size effects and book-to-market-value ratio effects indeed did exists in countries aside from the United States. Specifically, it is found that there are strong evidences that the market factors (as represented by statistically significance of beta coefficient) is highly significant in explaining the variation of cross section of average stock returns in all of the markets being investigated. However, the magnitude of significance of the other two factors, namely the firm size factor and book to market equity factor differ across the three countries, namely, Indonesia, Singapore and Taiwan. Nevertheless, both of the two factors are statistically significant in explaining the variation of cross section average stock returns in these countries under investigation. Overall, the authors able to show empirical evidences suggesting that the three factors models are true even outside the United States. The authors are of the opinion that the three factors do a reasonably satisfactory job in explaining the cross section average stock returns in the countries being investigated.
Then, Hearn (2010) conducted a study to investigate if firm size and liquidity are priced factors in some of the South Asian nations. The countries investigated include India, Pakistan, Bangladesh and Sri Lanka. He successfully found statistically significant empirical evidences that both firm size and liquidity are priced factors in all of the nations being investigated except in Sri Lanka. Similarly, Singh (2009) had conducted study to investigate if variables such as firm size, book-to-market-value ratio and others are being priced in stock returns in India. It is found that there are some empirical evidences supporting that both firm size and book-to-market-value ratio are indeed a priced factors in explaining cross sectional stock returns in India. Apart from that, Dempsey (2010) also conducted a study to investigate if the book-to-market-value ratio is indeed being priced to stock returns in Australian stock markets. It is found that the book-to-market-value ratio is exhibiting strong statistical relationships with stock returns in Australia. This finding indicates that book-to-market-value ratio is indeed at least one of the priced risk factors in Australia. This study is similar to a study conducted by Kassimatis (2008) earlier. According to the study conducted by Kassimatis (2008), there are some degree of empirical evidences that both firm size as well as book-to-market-value ratio has statistically significant explanatory power in explaining the variation of stock returns in Australia.
Besides, there was also study investigating if Fama and French three factor model still stay valid in the context of United States in the recent years. Simlai (2009) conducted a study on New York Stock Exchange, American Stock exchange and National Association of Securities Dealers Automated Quotations, and found that both the firm size and book-to-market-value ratio are indeed able to capture the cross sectional average stock returns in United States, within the research period from 1926 to 2007. Then, Walid (2009) had also conducted research to investigate if firm characteristics such as firm size and book-to-market-value ratio can be used to explain cross sectional variation in stock returns in Japan in the recent years. The research period started from 2002 to 2007. In the study, it is found that there are empirical evidences supporting both firm size and book-to-market-value ratio as a statistically significant priced factor in explaining variation of stock returns in Japan.
However, not all studies on three factor model yield positive results. For example, according to a study by Nartea, Ward & Djajadikerta (2009), there are some mixed findings if Fama and French three factors model can be employed to explain the cross sectional average stock returns in New Zealand stock market. Although it is found that Fama and French three factor model has slight improvement over the explanatory power of CAPM, a large part of the variation in stock returns is still unexplained. Then, it is also discovered that book-to-market-value ratio is indeed a priced factor in New Zealand stock exchange, but that is not the case for firm size. According to a study conducted by Mirza & Afzal (2011), the performance of Fama and French three factor model on fifteen European countries are investigated. The researchers employed the exact techniques employed by Fama and French (1992) to in the design of their research. The research period start from January 2002 towards December 2006. However, the three factor model is found to be unable to explain any variation in cross sectional average stock returns in these countries (in a statistically significant manner). Precisely, the two factors, namely, firm size and book-to-market-value ratio, failed to explain the variation of stock returns in these countries. In a similar vein, Novak & Petr (2010) conducted a study to investigate several commonly used risk factors in explaining cross sectional average stock returns. These factors include market returns (i.e., beta), firm size, book-to-market-value ratio, and momentum factor (as proxy by short term historical stock returns). The authors found none of these factors able to explain the cross sectional average stock returns in Stockholm Stock Exchange (in a statistically significant manner). As such, the authors perceived that the previously documented statistically significant relationship of these factors, namely market returns, firm size, book-to-market-value ratio, and momentum factor, are contingent on the data sample used on the specific time period.
From another perspective, there are also studies on CAPM and three factors models based on the assumptions that risk change across time (technically speaking, risk is assumed to be inter-temporarily constant). For example, Howton and Peterson (1999) argued that CAPM is just a single period security pricing model, and such an assumption is not true, whereby beta may not be constant over time. According to the authors, the trading environment is dynamic, and thus, when the risk is assumed to be constant, the CAPM will not be able to completely describe the returns in real life. Thus, the authors argued that security pricing model should take the notion that risk will change over time into account, and researchers should allow for shifts of risks in the investigation on risk and returns on financial markets. Howton and Peterson (1999) conducted their studies to allow for risk shifts through time in different market as well as economic states. In their regression models, stock returns are studied in relation to beta, firm size, book-to-market equity, and earning to price ratios. It is found that risk premiums vary across market (i.e., across New York Stock Exchange: NYSE, American Stock Exchange: AMEX and National Association of Security Dealers Automated Quotation System: NASDAQ) and economic states. Worst, the authors also found that risk premiums for the variables under studied indeed varied during January and non-January months. Overall, their study contributes to the discussion of asset pricing by urging for the use of conditional models as the relationships of returns and explanatory variables changes across time. Moreover, the importance of famously investigated variables such as firm size, book to market value ratio, earning to price ratios and economic factors beta are found to vary between January and non-January months. All these are strong evidences suggesting that systematic factors may affect returns in a conditional manner. Such findings are also supported by Mei (1993). In their study, the multifactor model is tested using time-varying risk premiums (through a quasi-differencing approach) in United States. The authors found that the ‘firm size effect’ and ‘dividend yield effects’ indeed exists and being captured in the multifactor models employed in the study. Overall, their study provide further evidences that constant beta version of multifactor model will not be (satisfactory) in explaining the cross section of average stock returns.
Overall, it is discussed that the introduction of CAPM to modern finance theory had revolutionarized how the entire industry think about risk and return issues in financial market. However, the one dimensional nature of CAPM, that employed excess return on market portfolio as the risk factors to explain stock returns fail to explain the variation of stock returns around the world satisfactorily. In other words, empirical testing on CAPM does not yield meaningful results, and beta as the risk measures has been receiving a great deal of critiques from both scholars and practitioners. Researchers have been associating the poor empirical results as beta is only a one-dimensional measure of risk, while the risks pertaining to financial market should be multi-dimensional. Thus, multifactor models or frameworks are increasingly being employed by researchers to study the relationships of certain risk factors to stock returns in various contexts. There are many popular and theoretically sound multifactor models available. However, one of the most famous multifactor models is the three factor model developed by Fama and French (1992).
In this dissertation, Fama and French three factors model will be employed to test the cross sectional average stock returns in Kuala Lumpur Stock Exchange (KLSE) in Malaysia. Fama and French three factors model is selected because it is popular; and a lot of tests had been performed on the model in different countries around the world. In other words, empirical evidences, either supporting or not supporting the model are available in literature. In this chapter, the literature concerning Fama and French three factor model in different countries around the world, which include Pakistan, India, Pakistan, Bangladesh, Sri Lanka, Indonesia, Singapore, Taiwan, Australia, New Zealand, 15 European countries, Sweden, Japan and China are reviewed and presented. Generally speaking, the empirical evidences concerning Fama and French three factor model are not conclusive. The empirical evidences reviewed are arranged in Table 2.1 below. As presented in the table, those researchers findings empirical evidences supporting Fama and French three factor model include: Javid (2008); Singh (2009); Hearn (2010); Naughton and Veeraraghavan (2005); Walid (2009); Dempsey (2010); Kassimatis (2008); and Simlai (2009). Then, researchers found mixed empirical evidences on Fama and French three factor model are Drew, Naughton and Veeraraghavan (2005) and Nartea, Ward & Djajadikerta (2009). Lastly, there are studies indicating that Fama and French three factor model has no explanatory power in explaining variation of stock returns in certain countries. Example of such studies include: Mirza & Afzal (2011); and Novak & Petr (2010).
Obviously, there are much to be done to validate the accuracy or relevancy of Fama and French three factor model in other countries around the world. Malaysia is selected because none of the previous research is studying the effectiveness of the model in Malaysian context. The specific research procedures, assumptions, methodologies and steps will be outlined in the next chapter.